Question

I have a procedure which performs some task, but performs it in two slightly different ways depending on its input. Specifically it extends a suffix in a suffix tree; if the suffix ends at a node the case is simple, but when the suffix ends within a leaf edge some additional work is required. I mention this detail to explain why I have included these two behaviours in a single function (it mirrors “Rule 2” of Dan Gusfield’s description of Ukkonen’s algorithm for suffix tree construction: http://www.stanford.edu/~mjkay/gusfield.pdf).

Anyway, once the function completes this work, the caller needs to know which of the two cases were followed. I thought an enum would be a good way to share this information, as it makes the cases explicit (as opposed to arbitrarily mapping the cases to bools or ints).

TLDR: To share information with the caller, should I pass an enum by reference to this procedure or return an enum? I feel like passing an enum by reference is better because it avoids having a ‘function with side effects’ but would like to know if there is generally ‘right’ way of doing this. Or, alternatively, does this really suggest that I should be replacing my single procedure with two separate procedures?

Answer 1

In my opinion it makes more sense to return the enum, since it is the result from calling the function. Using a reference would make much more sense if you were passing a value to the function, and the function was supposed to modify that value.

You can see a similar example in insert method from unordered_map in the C++ standard library:

template <class P> 
std::pair<iterator, bool> insert( P&& value );

That method returns a boolean (within a pair) stating whether the value was inserted in the map or not. This actually a very similar case as the one you are dealing with.